4.5. Standalone Forcing

Users are strongly encouraged to run CICE in a coupled system (see Coupling With Other Climate Model Components) to improve quality of science. The standalone mode is best used for technical testing and only preliminary science testing. Several different input forcing datasets have been implemented over the history of CICE. Some have become obsolete, others have been supplanted by newer forcing data options, and others have been implemented by outside users and are not testable by the Consortium. The forcing code has generally not been maintained by the Consortium and only a subset of the code is tested by the Consortium.

The forcing implementation can be found in the file cicecore/cicedyn/general/ice_forcing.F90. As noted above, only a subset of the forcing modes are tested and supported. In many ways, the implemetation is fairly primitive, in part due to historical reasons and in part because standalone runs are discouraged for evaluating complex science. In general, most implementations use aspects of the following approach,

• Input files are organized by year. The underlying implementation provides for some flexibility and extensibility in filenames. For instance, JRA55 and JRA55do filenames can have syntax like [JRA55,JRA55do][_$grid]_03hr_forcing_$year.nc or [JRA55,JRA55do]_03hr_forcing[_$grid]_$year.nc, where [_$grid] is optional and may be present at one of two locations within the filename. This implementation exists to support the current naming conventions within the gx1, gx3, and tx1 JRA55 and JRA55do CICE_data directory structure automatically. See JRA55_files in ice_forcing.F90 for more details.- Namelist inputs fyear and ycycle specify the forcing year dataset.

• The forcing year is computed on the fly and is assumed to be cyclical over the forcing dataset length defined by ycycle.

• The namelist atm_data_dir specifies the full or partial path for the atmosphere input data files, and the namelist atm_data_type defines the atmospheric forcing mode (see forcing_nml in Tables of Namelist Options). Many of the forcing options are generated internally. For atmospheric forcing read from files, the directory structure and filenames depend on the grid and atm_data_type. Many details can be gleaned from the CICE_data directory and filenames as well as from the implementation in ice_forcing.F90. The primary atm_data_type forcing for gx1, gx3, and tx1 test grids are JRA55 and JRA55do. For those configurations, the atm_data_dir should be set to ${CICE_data_root}/forcing/${grid}/[JRA55,JRA55do] and the filenames should be of the form [JRA55,JRA55do]_${grid}_03hr_forcing${atm_data_version}_yyyy.nc where yyyy is the forcing year. Those files should be placed under atm_data_dir/8XDAILY. atm_data_version is a string defined in forcing_nml namelist that supports versioning of the forcing data. atm_data_version could be any string including the null string. It typically will be something like _yyyymmdd to indicate the date the forcing data was generated.

• The namelist ocn_data_dir specifies the directory of the ocean input data files and the namelist ocn_data_type defines the ocean forcing mode.

• The filenames follow a particular naming convention that is defined in the source code (ie. subroutine JRA55_files). The forcing year is typically found just before the .nc part of the filename and there are tools (subroutine file_year) to update the filename based on the model year and appropriate forcing year.

• The input data time axis is generally NOT read by the forcing subroutine. The forcing frequency is hardwired into the model and the file record number is computed based on the forcing frequency and model time. Mixing leap year input data and noleap model calendars (and vice versa) is not handled particularly gracefully. The CICE model does not read or check against the input data time axis.

• Data is read on the model grid, no spatial interpolation exists.

• Data is often time interpolated linearly between two input timestamps to the model time each model timestep.

In general, the following variables need to be defined by the forcing module,

From Atmosphere:

• zlvl = atmosphere level height (m)

• uatm = model grid i-direction wind velocity component (m/s)

• vatm = model grid j-direction wind velocity component (m/s)

• strax = model grid i-direction wind stress (N/m^2)

• stray = model grid j-direction wind stress (N/m^2)

• potT = air potential temperature (K)

• Tair = air temperature (K)

• Qa = specific humidity (kg/kg)

• rhoa = air density (kg/m^3)

• flw = incoming longwave radiation (W/m^2)

• fsw = incoming shortwave radiation (W/m^2)

• swvdr = sw down, visible, direct (W/m^2)

• swvdf = sw down, visible, diffuse (W/m^2)

• swidr = sw down, near IR, direct (W/m^2)

• swidf = sw down, near IR, diffuse (W/m^2)

• frain = rainfall rate (kg/m^2 s)

• fsnow = snowfall rate (kg/m^2 s)

From Ocean:

• uocn = ocean current, x-direction (m/s)

• vocn = ocean current, y-direction (m/s)

• ss_tltx = sea surface slope, x-direction (m/m)

• ss_tlty = sea surface slope, y-direction (m/m)

• hwater = water depth for basal stress calc (landfast ice)

• sss = sea surface salinity (ppt)

• sst = sea surface temperature (C)

• frzmlt = freezing/melting potential (W/m^2)

• frzmlt_init= frzmlt used in current time step (W/m^2)

• Tf = freezing temperature (C)

• qdp = deep ocean heat flux (W/m^2), negative upward

• hmix = mixed layer depth (m)

• daice_da= data assimilation concentration increment rate (concentration s-1)(only used in hadgem drivers)

All variables have reasonable but static defaults and these will be used in default mode.

To advance the forcing, the subroutines get_forcing_atmo and get_forcing_ocn are called each timestep from the step loop. That subroutine computes the forcing year (fyear), calls the appropriate forcing data method, and then calls prepare_forcing which converts the input data fields to model forcing fields.

4.5.1. JRA55 and JRA55do Atmosphere Forcing

The current default atmosphere forcing for gx3, gx1, and tx1 standalone grids for Consortium testing is the JRA55 forcing dataset [61]. The Consortium has released 5 years of forcing data, 2005-2009 for gx3, gx1, and tx1 grids. Each year is a separate file and the dataset is on a gregorian time axis which includes leap days.

Schematic of JRA55 CICE forcing file generation.

The forcing is read and interpolated in subroutine JRA55_data. In particular, air temperature (airtmp), east and north wind speed (wndewd and wndnwd), specific humidity (spchmd), incoming short and longwave radiation (glbrad and dswsfc), and precipitation (ttlpcp) are read from the input files. The JRA55 reanalysis is run with updated initial conditions every 6 hours and output is written every 3 hours. The four state fields (air temperature, winds, specific humidity) are instantaneous data, while the three flux fields (radition, precipitation) are 3 hour averages. In the JRA55 forcing files provided by the Consortium, the time defined for 3 hour average fields is shifted 3 hours to the start time of the 3 hour interval. NOTE that this is different from the implementation on the original JRA55 files and also different from how models normally define time on an accumulated/averaged field. This is all shown schematically in Figure Schematic of JRA55 CICE forcing file generation..

The state fields are linearly time interpolated between input timestamps while the flux fields are read and held constant during each 3 hour model period. The forcing frequency is hardwired to 3 hours in the implementation, and the record number is computed based on the time of the current model year. Time interpolation coefficients are computed in the JRA55_data subroutine.

The forcing data is converted to model inputs in the subroutine prepare_forcing called in get_forcing_atmo. To clarify, the JRA55 input data includes

• uatm = T-cell centered, model grid i-direction wind velocity component (m/s)

• vatm = T-cell-centered, model grid j-direction wind velocity component (m/s)

• Tair = air temperature (K)

• Qa = specific humidity (kg/kg)

• flw = incoming longwave radiation (W/m^2)

• fsw = incoming shortwave radiation (W/m^2)

• fsnow = snowfall rate (kg/m^2 s)

and model forcing inputs are derived from those fields and the defaults.

Because the input files are on the gregorian time axis, the model can run with the regular 365 day (noleap) calendar, but in that case, the Feb 29 input data will be used on March 1, and all data after March 1 will be shifted one day. December 31 in leap years will be skipped when running with a CICE calendar with no leap days.

JRA55do forcing is also provided by the Consortium in the same format and scheme. The JRA55do dataset is more focused on forcing for ocean and ice models, but provides a very similar climate as the JRA55 forcing. To switch to JRA55do, set the namelist atm_data_type to JRA55do and populate the input data directory with the JRA55do dataset provided by the Consortium.

4.5.2. NCAR Atmosphere Forcing

The NCAR atmospheric forcing was used in earlier standalone runs on the gx3 grid, and the Consortium continues to do some limited testing with this forcing dataset. Monthly average data for fsw, cldf, fsnow are read. 6-hourly data for Tair, uatm, vatm, rhoa, and Qa are also read. Users are encouraged to switch to the JRA55 (see JRA55 and JRA55do Atmosphere Forcing) dataset. This atmosphere forcing dataset may be deprecated in the future.

4.5.3. Default Atmosphere Forcing

The default atmosphere forcing option sets the atmosphere forcing internally. No files are read. Values for forcing fields are defined at initialization in subroutine init_coupler_flux and held constant thereafter. Different conditions can be specified thru the default_season namelist variable.

4.5.4. Box2001 Atmosphere Forcing

The box2001 forcing dataset in generated internally. No files are read. The dataset is used to test an idealized box case as defined in [20].

4.5.5. Other Atmosphere Forcing

There are a few other atmospheric forcing modes, as defined by atm_data_type, but they are not tested by the Consortium on a regular basis.

4.5.6. Default Ocean Forcing

The default ocean setting is the standard setting used in standalone CICE runs. In this mode, the sea surface salinity is set to 34 ppt and the sea surface temperature is set to the freezing temperature at all grid points and held constant unless the mixed layer parameterization is turned on, in which case the SST evolves. Other ocean coupling fields are set to zero. No files are read.

4.5.7. Other Ocean Forcing

There are a few other ocean forcing modes, as defined by ocn_data_type, but they are not tested by the Consortium on a regular basis.